Revolutionising Treasury: How Companies Are Harnessing the Power of AI

The third HSBC Treasury Hangout was held in Dubai in June 2025, where treasurers from leading companies in the region discussed how Artificial Intelligence (AI) has the power to transform the treasury function, the current use cases, the hurdles to deeper AI adoption, and what treasury teams can do to get AI-ready.

The discussion was led by Chet Patel, an AI expert from HSBC’s Treasury Solutions Group, who presented a number of findings from interactions with HSBC treasury clients around the world and their AI journeys.

The key takeaway was that the integration of AI in treasury operations is a growing reality even if the AI journey for many companies is just beginning.

Where companies are on their AI journey

We began by asking the participants in a real time survey where they were in their AI journey in treasury. The survey results indicated the following:

• Just exploring: 68%
• Piloting specific use cases: 18%
• Not currently considering AI: 9%
• Actively implementing solutions: 5%
• Fully embedded in operations: 0%

The Hangout participants, similar to other HSBC clients, are largely in the exploratory phase when it comes to AI in treasury. Some are testing limited use cases on a small scale, but very few have fully integrated AI into their operations.

We also asked the participants how confident they were in their understanding of AI. The survey results indicated the following:

• Learning phase: 83%
• Very confident: 6%
• Somewhat confident: 6%
• Not confident: 6%

The attendees noted that given the critical financial role of the treasury function, there is a strong desire for the technology to become more reliable and secure before widespread adoption.

Wider HSBC research indicates that around a quarter of finance leaders are currently exploring AI tools. But the slow development of treasury-specific AI tools that encompass all the activities undertaken by the treasury function is slowing broader adoption.

Despite this caution, the underlying sentiment of the discussion was largely positive on the impact that AI could have on treasury.

Specific use cases of AI in treasury operations

We then asked the participants in a real time survey where they saw the most immediate potential for AI in Treasury. The survey results indicated the following:

• Fraud detection: 33%
• Cash forecasting: 25%
• Risk management: 17%
• Liquidity management: 8%
• Payments optimisation: 8%
• Other: 8%

Fraud detection: AI can continuously monitor bank accounts and payment types to flag suspicious activity outside of normal parameters, providing an early warning system for potential fraud. This builds on existing systems, controls, and daily reconciliations. Fraudulent invoices with changed bank details are areas where AI can detect anomalies more easily than manual processes.

Robotic Process Automation (RPA): RPA, a foundational element of AI, is a popular starting point for automating repetitive, rule-based tasks. The treasurers at the Hangout reported using RPA to streamline processes, such as bank account opening, which involve filling out forms with existing data.

Automation of administrative tasks: General-purpose AI tools, such as Microsoft Co-pilot and Google Gemini, are being used for automating mundane administrative tasks that consume a lot of treasury time. These tools are often already authorised within organisations, making their adoption in treasury relatively straightforward. These tasks include document translation, KYC processes, and data cleaning and organisation. One treasurer explained: “When I look at AI I am focusing on tasks I don't like doing as a treasurer, such as KYC or bank account openings, where the data is there but just needs to be populated into a form.”

Currency management strategies: AI can consume, analyse and present market information to suggest currency conversion strategies, offering multiple options to treasurers in a matter of minutes, a task that normally takes much longer.

Improved cash flow forecasting: By analysing large volumes of fragmented data, AI could improve the accuracy of cash flow predictions. However, one treasurer noted: “With cash flow forecasting, even when the data is there, I would want to check it to make sure it is accurate, as it is further along in the journey.”

One important distinction to emerge from the discussion was between traditional AI (often machine learning, focused on rule-based recognition) and generative AI (capable of creating new concepts and content). While machine learning has been used in treasury management systems for more than a decade, generative AI is still in its infancy in treasury.

Obstacles to AI adoption

We asked the participants in a third real time survey to highlight the major obstacles they faced in the adoption of AI in the work. The survey revealed the following results:

• Data quality or availability: 44%
• Unclear ROI or business case: 19%
• Integration with legacy systems: 19%
• Lack of internal expertise: 13%
• Regulatory/compliance concerns: 6%
• Budget constraints: 0%

Data quality and fragmentation: A fundamental challenge is the availability of clean, consistent, and centralised data. Many organisations struggle with fragmented data across disparate CRM, ERP, and TMS systems that don't communicate easily. This hinders AI's ability to provide insights. One treasurer at the Hangout noted that: "we have been using AI for a long time in our business. The challenge for treasury is how to get the different tools to talk to each other. Until these systems can talk to each other or you have undertaken a full digital transformation, you won’t get the best out of AI."

Trust and explainability: Treasurers need clear and explainable AI processes to validate results, especially for critical functions such as cash flow forecasting.

Skills gap: Treasury teams need to be augmented with strategists and technologists who possess expertise in both finance and technology, rather than relying solely on IT departments.

Rationale for investment: Securing investment in treasury-specific AI projects can be challenging. AI investments need to be able to demonstrate a clear and quantifiable return.

Resistance to change: Many treasury professionals rely heavily on traditional tools like Excel and may resist transitioning to more structured, AI-powered systems for fear of losing their jobs.

AI tools maturity: Fully-fledged mature AI tools (featuring all aspects of AI such as machine learning and generative AI), which perform a multitude of tasks are not yet available. It is unlikely that there will be any time soon a single tool that does everything that treasurers will need.

Incorporating AI in treasury operations

One treasurer noted how their company is exploring different ways of incorporating AI into their treasury operations. “We have hired a quant as an AI specialist and we are in the exploring phase of adopting AI. A lot of our mundane tasks, such a covenant trackers, have been automated. But there is more value in getting AI to do the harder tasks such as cash flow forecasting.”

We asked the participants to prioritise their AI activities over the next one to two years. The results of the survey indicated the following:

• Automate manual processes: 58%
• Enhance decision-making: 25%
• Increase accuracy of forecasts: 17%
• Improve fraud detection/prevention: 0%

To successfully reap the benefits of AI, the participants in the Hangout discussed a number of different strategies to incorporate AI into the treasury function:

Establishing a robust digital environment: AI thrives on digital data. All manual or paper-based processes must be digitised first, creating a digitally-enabled treasury environment. A comprehensive data strategy is crucial.

Integrate data and systems: Ensure that disparate systems such as ERP, TMS, CRM, banking sources, and market data are interconnected, ideally via APIs, to provide AI with comprehensive real-time data.

Prioritise AI governance and explainability: Integrate governance into AI processes from the start. Being able to explain AI results are paramount to building trust.

Developing a strong people strategy: Develop treasury teams that are a mix of finance professionals, strategists, and technologists. These individuals must understand both treasury operations and the technical aspects of AI. Focus how AI can free up capacity for more strategic activities.

Treat AI like a junior team member: Set realistic expectations. AI tools are still learning and won't solve every problem right away. They require time, energy, and human supervision to improve, just like a junior employee.

Human-in-the-loop approach: Maintain human oversight to compare AI results with expert judgment. This is vital for identifying data gaps or external factors not captured by AI. Human intervention will remain at the heart of complex decision-making.

Start with automation of mundane tasks: Start with readily available AI tools like Co-pilot and Gemini to automate repetitive, high-volume, low-complexity tasks such as KYC processes, bank account openings, and data cleansing.

Expand the scope of investment: When looking at AI investment, frame AI initiatives beyond treasury. Incorporate benefits for other functions to demonstrate broader organisational value and achieve better returns for the same level of investment.

Engage with suppliers critically: When discussing AI capabilities with suppliers, challenge their claims. Ask specific questions about whether your tools use machine learning or generative AI and how they address data integration and explainability.

While AI's full potential in treasury is still being developed, it already has the ability to automate mundane tasks, improve data visibility, and deliver predictive insights. By strategically addressing data quality, nurturing talent, establishing strong governance, and adopting a phased adoption approach, treasuries can progressively leverage AI to become smarter, more data-driven, and adaptable.

The next HSBC Treasury Hangout will be held in the Autumn and will focus on the future of treasury. We thank all the participants for their time and their insights.

Read the “Navigating the AI Wave:The Future of AI in Treasury” article where we explores how AI has the potential to redefine treasury operations—helping corporates navigate complexity, enhance decision-making, and unlock meaningful cost efficiencies.